Researchers developed a new deep learning method that boosts microplastic classification accuracy using the attention mechanism, outperforming traditional algorithms. The method achieves a classification accuracy of up to 98%, optimizing the model's performance and extracting more spectral features.
Researchers developed a novel immobilization strategy for surface plasmon resonance (SPR) assays of membrane proteins, effectively addressing technical constraints. The approach allows stable and specific capture of membrane proteins on sensor chips, enabling precise quantification of binding kinetics.
Researchers have discovered a single organic molecule can induce the Kondo effect, challenging long-held beliefs that it requires vast metallic electrons. The cobalt phthalocyanine molecule creates a 'molecular Kondo box' by hybridizing with conduction electrons of an underlying metal substrate.
Researchers at Hefei Institutes of Physical Science developed an ultra-compact excimer laser, eliminating mechanical gas pumps and reducing system volume. The laser achieved outstanding energy stability with a pulse energy exceeding 2 mJ while maintaining energy stability.
A new deep learning model, DSTMA-BLSTM, significantly enhances roadside air pollutant forecasting accuracy by 30% compared to conventional LSTM models. It achieves R² values exceeding 0.94 across major pollutants and demonstrates stable performance under complex traffic-meteorological coupling conditions.
A new method enhances precision and reliability of trace gas analysis in open-path infrared spectroscopic remote sensing. The integrated retrieval method combines VDL-DTCWT with NLLS fitting, significantly reducing background interference and improving pollutant concentration retrieval accuracy.
A research team at Hefei Institutes of Physical Science discovered a new mechanism for abnormal voltage attenuation in P2-type layered oxide cathodes. The study found that accumulation of unreduced molecular O2 leads to structural degradation and performance loss, highlighting the importance of bulk modification strategies.
A study improves marine aerosol remote sensing accuracy using multiangular polarimetry, which increases the degrees of freedom for signal (DFS) by at least 1.02 with added shortwave infrared measurements. This leads to the retrieval of one to two additional aerosol parameters and enhances understanding of columnar volume concentration,...
A team developed a bioinspired prussian blue/PNIPAM nanohybrid with spatiotemporally decoupled release characteristics, combining rapid and sustained release modes. The PAPP nanopesticides exhibited strong insecticidal activity while reducing harm to crops and non-target organisms.
Researchers developed a novel two-step plasma strategy to create surface-functionalized silver/mesoporous silica composites with strong antibacterial activity and positive surface charge. The modified materials showed improved interaction with bacterial membranes and effective suppression of E. coli infections.
Researchers developed an innovative approach to create dense dislocations in brittle superconductors, leading to a five-fold increase in critical current density. The technique utilizes asymmetric stress fields and scalable extrusion technology to induce localized lattice slip and twisting.
Researchers developed a novel robotic joint for heavy-duty robotics, achieving an ultra-high reduction ratio and delivering torque with low backlash. A deep reinforcement learning-based method also solved the problem of precise peg-in-hole assembly in radiation environments.
Researchers successfully engineered liquid-liquid phase separation-driven membraneless organelles (MLOs) in Corynebacterium glutamicum, resulting in a 2.43-fold increase in indigoidine titer. The technology enables spatial isolation of antimicrobial peptides, reducing toxicity and enabling successful expression.
Researchers developed an integrated method to accurately measure hemoglobin levels using near-infrared spectroscopy, overcoming limitations caused by strong water absorption and sample scattering. The approach significantly reduced background noise and achieved high accuracy, making it a reliable tool for non-invasive blood analysis.
Researchers developed a high-symmetry gradient-doped Nd:YAG laser crystal that reduces temperature differences between the center and ends, resulting in higher output power and improved beam quality. The crystal showed a 14 W output with efficiency boosted by more than half, making it suitable for dual-end pumping configurations.
A research team has discovered VOCs that can serve as reliable markers for multi-cancer screening. Early tumor signals were detectable in urine at week 5, in odor at week 13, and in feces at week 17.
A two-year field experiment found that one-time nitrogen application increases both soil and canopy ammonia emissions in maize fields. Split nitrogen application is a more sustainable choice with lower NH3 emissions and better maize productivity.
A team of researchers developed a self-sustaining solar-powered system that produces green hydrogen by coupling photothermal atmospheric water harvesting with proton exchange membrane electrolysis. The system maintains stable performance even under low humidity conditions, reaching a hydrogen production rate of nearly 300 mL per hour.
A team developed a method using generative adversarial networks (GANs) to learn underlying patterns in real spectrograms and reconstruct species-specific vocal components. This approach captures hidden features of the acoustic space, enabling precise separation of target sound sources and effective removal of environmental noise.
Researchers develop natural electron donor–assisted healing and targeted surface reconstruction strategy to regenerate degraded LiFePO₄ cathode materials. The approach reduces harmful defects and improves the rate performance of revived cathodes.
A new method for semi-supervised medical image segmentation outperforms traditional approaches by learning unified boundary feature representations across labeled and unlabeled data. The approach achieves competitive performance on benchmark datasets with as little as 10% labeled data.
A novel deep learning framework, CTCAIT, has been developed to detect early symptoms of Parkinson's, Huntington's, and Wilson disease through speech analysis. The model achieved a detection accuracy of 92.06% on a Mandarin Chinese dataset, demonstrating strong cross-linguistic generalizability.
A novel wearable eye patch using fluorescence sensors detects lysozyme levels in tears for early eye disease detection. The sensor's sensitivity is high enough to monitor lysozyme at concentrations as low as 1.5 nanomolar.
Researchers developed AI systems to enhance fusion energy experiments, predicting disruptions and monitoring plasma states with high accuracy. The tools improved reactor safety by 94% and achieved a 96.7% success rate in recognizing plasma conditions.
Researchers have discovered a new class of X-type antiferromagnetic materials that exhibit sublattice-selective spin transport and unconventional magnetic dynamics. This breakthrough enables precise control of the Néel vector, a crucial operation for antiferromagnetic spintronic data writing.
Researchers developed a high mechanical durability hydrogel electrolyte using urea as a zincophilic solubilizer, allowing for stable Zn stripping/plating in a dendrite and passivation-free manner. The material enhances the battery's overall performance and retention over time.
Researchers have proposed a third prototype of antiferromagnetic tunnel junction, paving the way for faster and denser spintronic devices. The new design, which relies on interface effects rather than bulk properties, allows for strong spin polarization and high performance.
A multi-omics study identified a key RNA-driven network behind colorectal cancer progression and immune resistance. Restoring a specific long non-coding RNA (lncRNA) expression reactivated tumor suppressor genes and enhanced immune cell infiltration, promoting anti-tumor effects.
Researchers developed a novel electroenzymatic platform enabling efficient non-natural oxidation reactions with high enantiomeric excess, opening a new chapter in biocatalysis for asymmetric synthesis
IHMT-15130, a selective BMX inhibitor, demonstrates robust efficacy in preclinical models of cardiac hypertrophy by suppressing inflammation and reversing pathological heart muscle thickening. The compound offers a safer therapeutic window for treating this major cardiovascular disorder.
Researchers developed a novel model optimization algorithm named External Calibration-Assisted Screening (ECA) to enhance the prediction robustness of Near-Infrared Spectroscopy (NIRS) quantitative models. ECA rapidly adapts initial models to new detection environments by calibrating them with externally collected samples.
Researchers found that 7-tesla magnetic fields accelerate demagnetization by 60% while suppressing efficiency by 34%, enabling widespread ultrafast demagnetization in other materials. Elevated temperatures enhance regulatory effects, promising real-world applications in high-speed storage and logic devices.
Researchers developed novel PbWO4 filler-reinforced B4C/HDPE composites with tunable microstructures, achieving enhanced synergistic radiation shielding performance. The optimized composite achieved a 97.32% shielding rate against ²⁵²Cf neutrons and 76.43% against ¹³⁷Cs gamma photons.
A recent study has shown that large language models (LLMs) can accurately predict liver cancer treatment responses, offering a new path towards AI-powered precision medicine. The LLMs demonstrated predictive accuracy on par with senior doctors, while outperforming junior and mid-level clinicians in speed and accuracy.
Researchers have engineered high-affinity antibodies against malaria by incorporating domains from inhibitory receptors, allowing them to block the interaction between parasite proteins and host immune cells. This new strategy offers a promising approach for treating malaria and could lead to innovative antimalarial drug discovery.
A compact all-solid-state CW SLM laser with high frequency stability was developed using iodine-based frequency locking, advancing its application in atmospheric remote sensing and environmental monitoring. The laser achieves long-term frequency stability with a drift of 4 MHz over a continuous 7-hour period.
A new AI-powered model has been developed to predict surface ozone concentration in the North China Plain and Yangtze River Delta regions. The model achieves high prediction accuracy, with hit rates of 83% and R² ≥ 0.85, providing a clearer picture of how weather patterns drive ozone pollution.
A new AI-powered model, PCWS, accurately predicts lung motion with a mean prediction error of just 0.20 mm, eliminating the need for 4D imaging. This approach represents a promising step toward personalized, low-risk treatment planning in lung cancer care.
Researchers propose a novel strategy to improve all-solid-state ion-selective electrodes by designing transduction layer materials with high hydrophobicity and large capacitance. The study reveals that ion-selective membranes constrain capacitive material performance, hindering sensor reliability.
A new dual-spectroscopy technique, Surface Plasmon-Enhanced Dual Spectroscopy (SPEDS), has been developed to detect hazardous chemicals in complex environments. This approach combines SERS and P-DUS, achieving molecular-level specificity while maintaining real-time responsiveness.
A new fluorescent probe detects pyrethroids with high sensitivity and rapid results, visible within 10 seconds. The technology is portable and allows for easy detection using a smartphone.
A research team detects aircraft-induced atmospheric disturbances using a novel approach with active light sources, enabling long-range sensing of aircraft presence. The detection scheme captures sharp spatial variations in atmospheric density within disturbed regions.
A research team developed a compact dynamic cantilever magnetometer with a diameter of 22 mm, achieving a magnetic moment sensitivity on the order of 10⁻¹⁷ A·m². This innovation provides strong technical support for frontier research areas like low-dimensional magnetism and quantum states of matter.
Researchers have developed a potent CDK9 inhibitor to combat drug-resistant hematological malignancies. The IHMT-CDK9-24 compound effectively inhibits both wild-type and L156F mutant forms of the enzyme.
Researchers at Hefei Institutes of Physical Science developed advanced aerogel composites with high-temperature insulation and mechanical load-bearing capabilities. The composites achieved controllable fabrication of large-size samples and demonstrated strong potential for future thermal protection applications.
Researchers introduced a novel low-thermal-effect gradient-doped crystal to improve brightness of high-power end-pumped Nd:YAG lasers. The optimized crystal extended absorption length, smoothed thermal gradients, and achieved conversion efficiencies above fifty percent.
Researchers developed CoNi-MOF nanozymes with laccase-like activity using a gas-liquid interface dielectric barrier discharge (DBD) low-temperature plasma (LTP) technique, exhibiting high catalytic degradation of tetracycline. The novel approach offers an eco-friendly strategy for addressing antibiotic pollution in the environment.
Researchers developed a method to grow titanium dioxide nanorod arrays with controllable spacing, achieving high-performance solar cells. The team's findings offer a new toolkit for crafting nanostructures across clean energy and optoelectronics.
A research team discovered stage-specific roles of plant hormones in regulating axillary bud development in Eucommia ulmoides. The study identified key hormone-related gene modules and constructed a regulatory network illustrating hormone crosstalk during bud growth.
A research team at the Chinese Academy of Sciences has developed high-performance 3D-printed graphene composites with improved thermal and electrical conductivity. The composites achieved an in-plane thermal conductivity of 4.54 W/(m·K) and showed enhanced photothermal conversion efficiency.